-
Notifications
You must be signed in to change notification settings - Fork 0
/
mul256x256_256_64bitlimbs_wip.wasm32.wat
275 lines (220 loc) · 10.8 KB
/
mul256x256_256_64bitlimbs_wip.wasm32.wat
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
(module
(memory 1)
(global $mul64x64_128_hi (mut i64) (i64.const 0))
(export "test" (func $test))
;; a = 0xaf6ed231c2b3174026b9fd3c903cde23dca5f06a04da8d06355a814a872d1782
;; b = 0x8ab8a663f6ba95916fbee2af731a6cc455ed5388343c8d226bc7232f315c2561
;; out_expected = 0x5ddcdd124d5f7664eedd3c724d9502470144bbfdd2c467a49e2f0f591133b242
(data (;0;) (i32.const 64) "\82\17\2d\87\4a\81\5a\35\06\8d\da\04\6a\f0\a5\dc\23\de\3c\90\3c\fd\b9\26\40\17\b3\c2\31\d2\6e\af")
(data (;0;) (i32.const 96) "\61\25\5c\31\2f\23\c7\6b\22\8d\3c\34\88\53\ed\55\c4\6c\1a\73\af\e2\be\6f\91\95\ba\f6\63\a6\b8\8a")
;; this is a filler function for mul:64x64->128, should assemble to native instead
;; takes params a,b and returns a*b, with hi limb returned through global $mul64x64_128_hi
(func $mul64x64_128 (param i64 i64) (result i64)
(local i64 i64 i64 i64 i64 i64 i64 i64)
;; convert to 32-bit limbs
(local.set 2 (i32.wrap/i64 (local.get 0)) (i64.extend_i32_u)) ;; a_lo
(local.set 3 (i32.wrap/i64 (i64.shr_u (local.get 0) (i64.const 32)))) ;; a_hi
(local.set 4 (i32.wrap/i64 (local.get 1)) (i64.extend_i32_u)) ;; b_lo
(local.set 5 (i32.wrap/i64 (i64.shr_u (local.get 1) (i64.const 32)))) ;; b_hi
;; multiply
(local.set 6 (i64.mul (local.get 2) (local.get 4)))
(local.set 7 (i64.mul (local.get 2) (local.get 5)))
(local.set 8 (i64.mul (local.get 3) (local.get 4)))
(local.set 9 (i64.mul (local.get 3) (local.get 5)))
;; middle 64-bits
(local.set 0
(i64.add
(local.get 8)
(i64.add
(i64.shr_u (local.get 6) (i64.const 32))
(local.get 7)
)
)
)
;; add overflow to local 9, even if 0
(local.set 9
(i64.add
(local.get 9)
(i64.shl
(i64.add
(local.get 0)
(i64.extend_i32_u (i64.gt_u (local.get 8) (local.get 0)))
)
(i64.const 32))
)
)
;; get lower 64-bit limb
(i64.add (local.get 3) (local.get 1)) (local.set 1) ;; add to accumulator t1
(i64.add (local.get 2) (local.get 3)) (local.set 2) ;; add to overflow, even if add zero
(i64.add
(i64.shl (local.get 7) (i64.const 32))
(i64.shl (local.get 8) (i64.const 32))
)
)
(local.set 0
(i64.shl
(i64.add
(i64.and (local.get 7) (i64.const 0xffffffff))
(i64.and (local.get 8) (i64.const 0xffffffff))
)
(i64.const 32)
)
)
(local.set 2 ;; carry
(i64.add
(local.get 6)
)
)
(local.set 1
(i64.add
(local.get 9)
(i64.add
(i64.shr_u (local.get 7))
(i64.shr_u (local.get 8))
)
)
)
(local.set 3 (i64.mul (local.get 3) (local.get 5)))
;; set high limb
(local.get 0) ;; blah
(global.set $mul64x64_128_hi)
;; set low limb
(i64.const 0) ;; blah
)
(func $test (result i64)
(local i64) ;; this local is returned to show which limb failed
i64.const 0
i64.const 64
i64.const 96
call $mul256x256_256
;; compare output to expected
block
;; test limb 0
(i32.ne (i32.load (i32.const 0)) (i32.const 288600642))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 1
(i32.ne (i32.load (i32.const 4)) (i32.const 2653884249))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 2
(i32.ne (i32.load (i32.const 8)) (i32.const 3536086948))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 3
(i32.ne (i32.load (i32.const 12)) (i32.const 21281789))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 4
(i32.ne (i32.load (i32.const 16)) (i32.const 1301611079))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 5
(i32.ne (i32.load (i32.const 20)) (i32.const 4007476338))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 6
(i32.ne (i32.load (i32.const 24)) (i32.const 1298101860))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
;; test limb 7
(i32.ne (i32.load (i32.const 28)) (i32.const 1574755602))
br_if 0
(local.set 0 (i64.add (i64.const 1)(local.get 0))) ;;increment
end
local.get 0 ;;return index where failed, or num_limbs if it passed
;;(i64.load (i32.const 0))
;;(i64.and (i64.load (i32.const 8)) (i64.const 0xffffffff)) ;; lower 32-bits
)
(func $mul256x256_256 (param i64 i64 i64)
(local i64 i64 i64 i64 i64 i64 i64 i64 i64 i64 i64)
;; bring a,b into locals 6-13
(i64.load (i32.wrap/i64 (i64.add (local.get 1) (i64.const 0)))) (local.set 6)
(i64.load (i32.wrap/i64 (i64.add (local.get 1) (i64.const 8)))) (local.set 7)
(i64.load (i32.wrap/i64 (i64.add (local.get 1) (i64.const 16)))) (local.set 8)
(i64.load (i32.wrap/i64 (i64.add (local.get 1) (i64.const 24)))) (local.set 9)
(i64.load (i32.wrap/i64 (i64.add (local.get 2) (i64.const 0)))) (local.set 10)
(i64.load (i32.wrap/i64 (i64.add (local.get 2) (i64.const 8)))) (local.set 11)
(i64.load (i32.wrap/i64 (i64.add (local.get 2) (i64.const 16)))) (local.set 12)
(i64.load (i32.wrap/i64 (i64.add (local.get 2) (i64.const 24)))) (local.set 13)
;; zero things
(local.set 3 (i64.const 0))
;; done initializing, now let's compute something
;; a0*b0 + carry
;; a0*b0
(call $mul64x64_128 (local.get 6) (local.get 10)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(local.set 2) (local.get 5)
(local.set 1) (local.get 4)
;; write result for this limb, prepare for next limb
(i32.store (i32.wrap/i64 (i64.add (local.get 0) (i64.const 0))) (i32.wrap/i64 (local.get 1)) ) ;; write result for this 32-bit limb
(local.set 1) (local.get 2)
(local.set 2) (i64.const 0)
;; a0*b1 + a1*b0 + carry
;; a0*b1
(call $mul64x64_128 (local.get 6) (local.get 11)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
(i64.gt_u (local.get 4) (local.get 1)) (local.set 4 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 4) (local.get 5)) (local.set 5) ;; add carry, even if add zero
(i64.add (local.get 2) (local.get 5)) (local.set 2) ;; add high limb to accumulator r2
(i64.gt_u (local.get 5) (local.get 2)) (local.set 5 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 3) (local.get 5)) (local.set 3) ;; add carry, even if add zero
;; a1*b0
(call $mul64x64_128 (local.get 7) (local.get 10)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
(i64.gt_u (local.get 4) (local.get 1)) (local.set 4 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 4) (local.get 5)) (local.set 5) ;; add carry, even if add zero
(i64.add (local.get 2) (local.get 5)) (local.set 2) ;; add high limb to accumulator r2
(i64.gt_u (local.get 5) (local.get 2)) (local.set 5 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 3) (local.get 5)) (local.set 3) ;; add carry, even if add zero
;; write result for this limb, prepare for next limb
(i64.store (i32.wrap/i64 (i64.add (local.get 0) (i64.const 8))) (i32.wrap/i64 (local.get 1)) ) ;; write result for this 32-bit limb
(local.set 1) (local.get 2)
(local.set 2) (local.get 3)
(local.set 3) (i64.const 0)
;; a0*b2 + a1*b1 + a2*b0 + carry
;; a0*b2
(call $mul64x64_128 (local.get 6) (local.get 12)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
(i64.gt_u (local.get 4) (local.get 1)) (local.set 4 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 4) (local.get 5)) (local.set 5) ;; add carry, even if add zero
(i64.add (local.get 2) (local.get 5)) (local.set 2) ;; add high limb to accumulator r2
(i64.gt_u (local.get 5) (local.get 2)) (local.set 5 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 3) (local.get 5)) (local.set 3) ;; add carry, even if add zero
;; a1*b1
(call $mul64x64_128 (local.get 7) (local.get 11)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
(i64.gt_u (local.get 4) (local.get 1)) (local.set 4 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 4) (local.get 5)) (local.set 5) ;; add carry, even if add zero
(i64.add (local.get 2) (local.get 5)) (local.set 2) ;; add high limb to accumulator r2
(i64.gt_u (local.get 5) (local.get 2)) (local.set 5 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 3) (local.get 5)) (local.set 3) ;; add carry, even if add zero
;; a2*b0
(call $mul64x64_128 (local.get 8) (local.get 10)) (local.set 4) (local.set 5 (global.get 0)) ;; global has upper half
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
(i64.gt_u (local.get 4) (local.get 1)) (local.set 4 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 4) (local.get 5)) (local.set 5) ;; add carry, even if add zero
(i64.add (local.get 2) (local.get 5)) (local.set 2) ;; add high limb to accumulator r2
(i64.gt_u (local.get 5) (local.get 2)) (local.set 5 (i64.extend_i32_u)) ;; check for carry
(i64.add (local.get 3) (local.get 5)) (local.set 3) ;; add carry, even if add zero
;; write result for this limb, prepare for next limb
(i64.store (i32.wrap/i64 (i64.add (local.get 0) (i64.const 16))) (i32.wrap/i64 (local.get 1)) ) ;; write result for this 32-bit limb
(local.set 1) (local.get 2)
(local.set 2) (local.get 3)
(local.set 3) (i64.const 0)
;; a0*b3 + a1*b2 + a2*b1 + a3*b0 + carry
;; a0*b3
(mul (local.get 6) (local.get 13)) (local.set 4) ;; regular multiplication since overflow is not needed
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
;; a1*b2
(mul (local.get 7) (local.get 12)) (local.set 4) ;; regular multiplication since overflow is not needed
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
;; a2*b1
(mul (local.get 8) (local.get 11)) (local.set 4) ;; regular multiplication since overflow is not needed
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
;; a3*b0
(mul (local.get 9) (local.get 10)) (local.set 4) ;; regular multiplication since overflow is not needed
(i64.add (local.get 1) (local.get 4)) (local.set 1) ;; add to accumulator r1
;; write result for this limb, prepare for next limb
(i64.store (i32.wrap/i64 (i64.add (local.get 0) (i64.const 24))) (i32.wrap/i64 (local.get 1)) ) ;; write result for this 32-bit limb
)
)